diff --git a/Neural graph module/ngm.ipynb b/Neural graph module/ngm.ipynb
index e5d9403df82c271b501aa62a75189c618cccc3ec..448cf36375aeab9161dd776cc696f2c298e3753a 100644
--- a/Neural graph module/ngm.ipynb
+++ b/Neural graph module/ngm.ipynb
@@ -31,15 +31,7 @@
"cell_type": "code",
"execution_count": null,
"metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "cuda\n"
- ]
- }
- ],
+ "outputs": [],
"source": [
"# Use GPU if available\n",
"device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
@@ -49,7 +41,7 @@
},
{
"cell_type": "code",
- "execution_count": 20,
+ "execution_count": null,
"metadata": {},
"outputs": [],
"source": [
@@ -215,19 +207,83 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 15,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Beginning making batch\n"
+ ]
+ },
+ {
+ "name": "stderr",
+ "output_type": "stream",
+ "text": [
+ "100%|██████████| 408/408 [00:00<00:00, 616.32it/s]\n"
+ ]
+ },
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Finished with batches\n",
+ "Beginning making batch\n"
+ ]
+ },
+ {
+ "name": "stderr",
+ "output_type": "stream",
+ "text": [
+ "100%|██████████| 408/408 [00:00<00:00, 620.42it/s]"
+ ]
+ },
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Finished with batches\n"
+ ]
+ },
+ {
+ "name": "stderr",
+ "output_type": "stream",
+ "text": [
+ "\n"
+ ]
+ },
+ {
+ "ename": "ValueError",
+ "evalue": "Sum of input lengths does not equal the length of the input dataset!",
+ "output_type": "error",
+ "traceback": [
+ "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
+ "\u001b[1;31mValueError\u001b[0m Traceback (most recent call last)",
+ "\u001b[1;32mc:\\Users\\Albin\\Documents\\TDDE19\\codebase\\Neural graph module\\ngm.ipynb Cell 7\u001b[0m in \u001b[0;36m<cell line: 14>\u001b[1;34m()\u001b[0m\n\u001b[0;32m <a href='vscode-notebook-cell:/c%3A/Users/Albin/Documents/TDDE19/codebase/Neural%20graph%20module/ngm.ipynb#X11sZmlsZQ%3D%3D?line=8'>9</a>\u001b[0m inputs, attention_mask, correct_rels \u001b[39m=\u001b[39m make_batch()\n\u001b[0;32m <a href='vscode-notebook-cell:/c%3A/Users/Albin/Documents/TDDE19/codebase/Neural%20graph%20module/ngm.ipynb#X11sZmlsZQ%3D%3D?line=12'>13</a>\u001b[0m dataset \u001b[39m=\u001b[39m MyDataset(\u001b[39m*\u001b[39mmake_batch(), relations\u001b[39m=\u001b[39mrelations)\n\u001b[1;32m---> <a href='vscode-notebook-cell:/c%3A/Users/Albin/Documents/TDDE19/codebase/Neural%20graph%20module/ngm.ipynb#X11sZmlsZQ%3D%3D?line=13'>14</a>\u001b[0m split_data \u001b[39m=\u001b[39m random_split(dataset, [\u001b[39m0.8\u001b[39;49m, \u001b[39m0.2\u001b[39;49m], generator\u001b[39m=\u001b[39;49mtorch\u001b[39m.\u001b[39;49mGenerator()\u001b[39m.\u001b[39;49mmanual_seed(\u001b[39m42\u001b[39;49m))\n\u001b[0;32m <a href='vscode-notebook-cell:/c%3A/Users/Albin/Documents/TDDE19/codebase/Neural%20graph%20module/ngm.ipynb#X11sZmlsZQ%3D%3D?line=15'>16</a>\u001b[0m train_dataloader \u001b[39m=\u001b[39m DataLoader(train_set, batch_size\u001b[39m=\u001b[39m\u001b[39m1\u001b[39m, shuffle\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m)\n\u001b[0;32m <a href='vscode-notebook-cell:/c%3A/Users/Albin/Documents/TDDE19/codebase/Neural%20graph%20module/ngm.ipynb#X11sZmlsZQ%3D%3D?line=17'>18</a>\u001b[0m \u001b[39m#show first entry\u001b[39;00m\n",
+ "File \u001b[1;32mb:\\Programs\\Miniconda\\envs\\tdde19\\lib\\site-packages\\torch\\utils\\data\\dataset.py:311\u001b[0m, in \u001b[0;36mrandom_split\u001b[1;34m(dataset, lengths, generator)\u001b[0m\n\u001b[0;32m 309\u001b[0m \u001b[39m# Cannot verify that dataset is Sized\u001b[39;00m\n\u001b[0;32m 310\u001b[0m \u001b[39mif\u001b[39;00m \u001b[39msum\u001b[39m(lengths) \u001b[39m!=\u001b[39m \u001b[39mlen\u001b[39m(dataset): \u001b[39m# type: ignore[arg-type]\u001b[39;00m\n\u001b[1;32m--> 311\u001b[0m \u001b[39mraise\u001b[39;00m \u001b[39mValueError\u001b[39;00m(\u001b[39m\"\u001b[39m\u001b[39mSum of input lengths does not equal the length of the input dataset!\u001b[39m\u001b[39m\"\u001b[39m)\n\u001b[0;32m 313\u001b[0m indices \u001b[39m=\u001b[39m randperm(\u001b[39msum\u001b[39m(lengths), generator\u001b[39m=\u001b[39mgenerator)\u001b[39m.\u001b[39mtolist()\n\u001b[0;32m 314\u001b[0m \u001b[39mreturn\u001b[39;00m [Subset(dataset, indices[offset \u001b[39m-\u001b[39m length : offset]) \u001b[39mfor\u001b[39;00m offset, length \u001b[39min\u001b[39;00m \u001b[39mzip\u001b[39m(_accumulate(lengths), lengths)]\n",
+ "\u001b[1;31mValueError\u001b[0m: Sum of input lengths does not equal the length of the input dataset!"
+ ]
+ }
+ ],
"source": [
- "#Prepare data\n",
+ "from torch.utils.data import random_split\n",
"\n",
+ "#Prepare data\n",
"def open_json(file):\n",
" with open(file, \"r\") as f:\n",
" return json.load(f)\n",
"\n",
+ "relations = open_json(\"../data/relations-query-qald-9-linked.json\")\n",
+ "inputs, attention_mask, correct_rels = make_batch()\n",
+ "\n",
+ "# relations = open_json(\"../data/relations-lcquad-without-http-train-linked.json\")\n",
+ "# train_set = MyDataset(*make_batch(), relations=relations)\n",
+ "\n",
+ "\n",
+ "dataset = MyDataset(*make_batch(), relations=relations)\n",
+ "split_data = random_split(dataset, [0.8, 0.2], generator=torch.Generator().manual_seed(42))\n",
"\n",
- "relations = open_json(\"../data/relations-lcquad-without-http-train-linked.json\")\n",
- "train_set = MyDataset(*make_batch(), relations=relations)\n",
"train_dataloader = DataLoader(train_set, batch_size=1, shuffle=True)\n",
"\n",
"#show first entry\n",
@@ -258,8 +314,10 @@
"epoch = 500\n",
"batch_size = 64\n",
"train_dataloader = DataLoader(train_set, batch_size=batch_size, shuffle=True)\n",
+ "valid_dataloader = DataLoader(valid_set, batch_size=batch_size, shuffle=True)\n",
"for e in range(epoch):\n",
- " epoch_loss = 0\n",
+ " train_loss_epoch = 0\n",
+ " valid_loss_epoch = 0\n",
" for i_batch, sample_batched in enumerate(train_dataloader):\n",
" optimizer.zero_grad()\n",
" train = sample_batched[0]\n",
@@ -273,9 +331,20 @@
" # backward and optimize\n",
" loss.backward()\n",
" optimizer.step()\n",
- " epoch_loss = epoch_loss + loss.item()\n",
+ " train_loss_epoch = train_loss_epoch + loss.item()\n",
+ "\n",
+ " for i_batch, sample_batched in enumerate(valid_dataloader):\n",
+ " valid = sample_batched[0]\n",
+ " valid_mask = sample_batched[1]\n",
+ " label_index = sample_batched[2].to(device)\n",
+ " \n",
+ " # Forward pass\n",
+ " output = model(train, train_mask)\n",
+ " loss = criterion(output, label_index)\n",
+ "\n",
+ " valid_loss_epoch = valid_loss_epoch + loss.item()\n",
"\n",
- " print(e+1, epoch_loss / len(sample_batched))"
+ " print(e+1, \"Train\", valid_loss_epoch/len(sample_batched), \", Valid \", valid_loss_epoch/len(sample_batched))"
]
},
{
@@ -321,7 +390,7 @@
],
"metadata": {
"kernelspec": {
- "display_name": "Python 3.9.11 64-bit",
+ "display_name": "Python 3.10.4 ('tdde19')",
"language": "python",
"name": "python3"
},
@@ -335,12 +404,12 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
- "version": "3.9.11"
+ "version": "3.10.4"
},
"orig_nbformat": 4,
"vscode": {
"interpreter": {
- "hash": "64e7cd3b4b88defe39dd61a4584920400d6beb2615ab2244e340c2e20eecdfe9"
+ "hash": "8e4aa0e1a1e15de86146661edda0b2884b54582522f7ff2b916774ba6b8accb1"
}
}
},